ObjectivesNon-persistence may be a significant barrier to the use of metformin. Our objective was to assess reasons for metformin non-persistence, and whether initial metformin dosing or use of extended release (ER) formulations affect persistence to metformin therapy.DesignRetrospective cohort study.SettingElectronic health record data from a network of urban academic practices.ParticipantsThe cohort was restricted to individuals receiving a metformin prescription between 2009/1/1 and 2015/9/31, under care for at least 6 months before the first prescription of metformin. The cohort was further restricted to patients with no evidence of any antihyperglycaemic agent use prior to the index date, an haemoglobin A1c measured within 1 month prior to or 1 week after the index date, at least 6 months of follow-up, and with the initial metformin prescription originating in either a general medicine or endocrinology clinic.Primary and secondary outcome measuresThe primary outcome measure was early non-persistence, as defined by the absence of further prescriptions for metformin after the first 90 days of follow-up.ResultsThe final cohort consisted of 1259 eligible individuals. The overall rate of early non-persistence was 20.3%. Initial use of ER and low starting dose metformin were associated with significantly lower rates of reported side effects and non-persistence, but after multivariable analysis, only use of low starting doses was independently associated with improved persistence (adjusted OR 0.54, 95% CI 0.37 to 0.76, for comparison of 500 mg daily dose or less to all higher doses).ConclusionsThese data support the routine prescribing of low starting doses of metformin as a tool to improve persistence. In this study setting, many providers routinely used ER metformin as an initial treatment; while this practice may have benefits, it deserves more rigorous study to assess whether increased costs are justified.
Pediatric BMI changes during COVID-19 pandemic: An electronic health record-based retrospective cohort study
